How to avoid creating a pushy bot

(Image credit: Image Credit: Geralt / Pixabay)

As AI technology matures, it’s the general consensus that customer service bots will move from reactive, order driven tools to proactive, sentient “beings” which understand what we’re looking for before we do. But as companies look to make the step change, they could end up making mistakes – and we could see the rise of the pushy bot, annoying us more than being on hold for 2 hours waiting to speak to an agent.

We spoke to bot expert Ryan Lester, Director of Customer Engagement Technologies at LogMeIn to find out how to avoid creating an annoying bot, but instead how to use AI to create a helpful, proactive digital assistant.

  • What is the technology behind producing a personalised ‘concierge’ customer service bot?

Proactive customer service chatbots use a combination of AI and rules-based algorithms to make decisions based on customer actions, data and previous interactions.  

The chatbot analyses that information to determine how they can interact in a way that is relevant to that particular customer.  For example, a bot may know that a customer lives in Brighton and every year at this time that customer looks to buy a new swimsuit.  

The chatbot can make relevant product recommendations based on the style and sizes the customer has purchased in the past, offer local stores where it is in stock and they can get it today.  

  • How do bot creators know when is the “right time” to introduce a bot to a customer? 

By using predictive analytics based on behavior, the bot can “predict” when a customer needs help.  If they are comparing two products and keep toggling back and forth, a bot may be introduced to help them make the purchasing decision.  If the bot knows that this particular customer only makes purchases when they have a discount to use, it may offer a discount to incentivise the customer to close the deal.  The idea is not for the bot to be annoying, but helpful. 

  • How do businesses ensure the bot does not become too pushy?

Proactive bots can be great if they offer value to a customer, so companies should be wary of implementing bots that are generic. Proactive is most valuable if it can take into account context of what’s learned from the customer journey to add value to that particular customer.  If it can’t do that, then it’s a pop-up which can come across as annoying. 

  • Will proactive AI mean an end to human customer service? How can they best work together?

No, this is a common misconception.  

Using chatbots for customer service should not be viewed as a direct replacement for the work human agents carry out. Bots and humans need to work hand in hand to create a great customer experience.  There are some things that bots are good at – those repetitive, simple queries like return policies or inventory questions and others that require a human touch.  

As companies today look to deploy chatbots, they need to be intelligent enough to understand when they can’t resolve a query, and escalate it seamlessly to a human agent including the context of the previous conversation. 

Ultimately, chatbots provide companies with the ability to better leverage their most valuable resources – their employees – not replace them.

  • We all remember Clippy, the Microsoft paperclip. “I see you are writing a letter. Can I help?” How is today’s proactive bot more advanced than the technology behind this? 

The problem with Clippy was that it was a high-interaction, low value assistant.  

Today’s technology is able to gather and process information on a customer and provide a deeper level of assistance.  These chatbots are designed to be additive to the journey and address friction points, but not be infuriating through the whole journey.  

In a retail scenario, a technology like Clippy would follow you around the entire store asking if it could help you even after you said you were “just browsing”.  Whereas today’s bot would not only be able to ascertain that you were buying a dress, but would able to tell where those pants were in stock, the return policy, a sweater you might like to go with those pants, if the fit had changed from previous pant purchases you made in the past, etc. 

  • When will customer bots be able to sense human emotion in customer responses?

Bots can already gleam the emotional mindset of a customer from their tone and choice of language. 

For example, if a customer is swearing or using words like annoyed or angry, it should be able to indicate that the customer is not in the best mood and decide the next course of action, whether that involves redirecting the customer, or escalating the issue to a human agent.  

AI will continue to mature it will be able to better understand human emotion and sentiment, but human agents will still be the best source to provide the highest level of empathy.